We’re talking 30+ years in most cases though (often a lot more than that) have amps built in the 90’s that I wouldn’t recap yet. But yes eventually they need to be replaced.
The bigger replacement issue in vintage electronics though is if the layers of the transformer core start to separate, that can often be not worthwhile even for expensive pieces.
1 Like
I leave my May and Spring 2 on all the time, behind ZeroSurge surge protectors.
2 Likes
Same here…I leave the May on. I do turn off the display so it’s not lit 24/7.
5 Likes
I left my Wavedream Signature on since I got it 2 years ago. Only turned off when I’m out of town. I think you’ll be fine.
2 Likes
Thanks all, I appreciate the input. 
On at all time is the rec from Kitsune, I asked. Takes about 24-48 hours to thermally stabilize from my experience when starting from cold. It does put out a fair bit of heat.
4 Likes
I’m waiting in the queue for a Holo Spring 3 KTE and Bliss KTE. I’ve been peripherally aware of i^s but never had a kit with that capability. What are the typical use cases for using i^s hdmi with Holo Audio Spring 3 DAC? My current chain uses a JCAT USB XE card (LPS powered) output to a Chord DAVE DAC that I sold to fund the Spring. I’ve never detected any shortcomings with my chain that I couldn’t address but I’m mildly curious to learn if there are sonic benefits for using i^s for my 2 channel audio. My library consists of PGGB 24-bit 706.6/768 kHz music files so I’m not sure whether this tech handles PCM of this sample rate better than USB. This is an opportunity for me to learn something new! After all I’m desperate to pass time until my Bliss KTE and Spring 3 arrive next month!
4 Likes
First, welcome to the forum!
I’m not an expert, but my understanding is that I2S splits the signal so that the time domain information is NOT transmitted, thus allowing the Spring 3 to control the time domain information with its higher end clock. So, you would feed a streamer or CD transport I2S out to Spring 3’s I2S in and theoretically achieve a more coherent analog result. In essence, it lets you leverage your high end dac to do more of the work to achieve the best possible result.
3 Likes
I2S like SPDIF/AES carries the clock, but unlike SPDIF and AES, the device doesn’t have to recover the clock from the logic signal because it’s sent independently.
So all 3 are usually dependent of the quality of the clock in the source.
There is nothing to stop a device “cleaning” up that clock using a PLL or similar.
USB on the other hand the DAC is entirely responsible for generating the clock, in fact there are many DAC’s that will internally generate an SPDIF or I2S signal from USB sources.
Theoretically the separate clock could make I2S better from a jitter standpoint but there is a lot more than just the connection involved in that.
4 Likes
You’ll need a DDC like the Holo Red to convert from USB to I2S.
According to Goldensound’s review, the Red I2S output supports up to 768khz PCM and DSD512.
Depending on how noisy your PC is, you may find that a DDC will clean up your signal to the DAC even more. I also own a JCAT USB XE (LPS powered) and noticed that the sound quality got subjectively better (i.e. changed in a way that I liked 
) when I put a Mutec DDC in my chain, to convert the signal from USB to SPDIF.
4 Likes
Thanks @chrisnyc75 for replying. Your description helped me understand the potential value of i2s. I like to have a compelling reason before adding another device in my chain but it makes sense to let the clock in the Spring manage the time domain.
1 Like
I2S is the “native” interface for most DAC chips. As a result, most discrete DACs (i.e. R2R/multi-bit and true 1-bit DSD) also use a native I2S interface … as that lets you interface to them via standard input boards and controllers.
For example, in most DACs using XMOS-based inputs, ALL the inputs come through the XMOS board and then that board simply outputs I2S to the converter stage. This allows that single XMOS board to handle all ASRC, up/down/resampling, MQA and other processing.
I2S was designed as a VERY short-range interface. The original design expected a connection length of <= 100mm (a bit over 4 inches). Using LVDS (as with your typical I2S interface on a DAC via HDMI connectors*) you can extend that, safely, significantly (though shorter is still better).
The principle benefit of I2S is that, excepting USB, it is the only common Transport->DAC protocol that can handle bitrates over 192 kHz (or rates higher than DSD64 or 1x DSD).
USB implementations, especially coming off “high-end” PCs can be extremely noisy (pollution on the GND/VCC lines) and poorly implemented USB receivers can allow that noise to get to the analog circuits in the DAC.
This was a much bigger issue a few years ago; many quality DACs now simply isolate the USB input and don’t allow the USB power lines a path beyond the USB receiver.
When I tried the original Holo Audio Spring DAC L3/KTE back in, oh, 2016, I noted that its USB input wasn’t very good (regardless of what trickery you put in front of it), and that I would really only use it with a DDC to bypass USB via I2S. Tim, of Kitsune HiFi/Holo Audio USA said much the same thing.
In contrast I didn’t find a reliable difference comparing either Spring 3 DAC or the May DAC via USB or I2S. It might be that you need a DDC on the level of the Holo Audio RED to expose that difference, or those DACs respective USB inputs may be good enough that it doesn’t matter.
Anyway, I’d try the DAC without the DDC/I2S first … and if you’re not thrilled, you can always add a DDC and try I2S later.
*It uses HDMI cabling/connectors, but it is NOT an HDMI signal.
9 Likes
Thanks Torq, @Polygonhell @PaisleyUnderground I have a much better understanding of i2S now. Torq, I was thinking along the lines you suggested of letting the new kit settle in for a month until the sound is more fully revealed and then address any perceived deficiencies. PaisleyUnderground thanks for the link to Goldensound’s Red review, I had seen it before but didn’t read it. It does look like it tests well. I like how flexible it is between streamer and a DDC, I like options.
2 Likes
When I tried the original Holo Audio Spring DAC L3/KTE back in, oh, 2016, I noted that its USB input wasn’t very good (regardless of what trickery you put in front of it), and that I would really only use it with a DDC to bypass USB via I2S. Tim, of Kitsune HiFi/Holo Audio USA said much the same thing.
In contrast I didn’t find a reliable difference comparing either Spring 3 DAC or the May DAC via USB or I2S.
Do you know if the Cyan 2 has the bad old USB implementation or the good new one? Or something else?
1 Like
Are there any Holo Audio dac owners that use Remastero’s PGGB remastering software? I’m on the PGGB beta crew and my library currently consists of 24 bit 16fs wav files. I’m awaiting delivery of a Spring 3 KTE as a replacement for my Chord DAVE DAC.
If you use PGGB, what output bit depth and sample rare are you using for Red Book recordings? The reason I ask is that I’ve read that R2R dacs have resistor non linearity and it’s very hard to maintain resistor tolerance to 24 bits, it’s just better to not use the last few bits. So I’m thinking the PGGB bit depth output may be better if set in the range 19-22 bits.
It’s a completely different implementation on the CYAN 2; but I can’t tell you how it performs vs. any other model in the line-up as I’ve not heard one.
As far as I can tell, the CYAN 2’s USB board is unique in the Holo Audio line-up in that it uses an Altera MAX II CLPD instead of an XMOS-based implementation. Whether this is a new overall direction (since the CYAN 2 is the latest DAC-product from Holo Audio), or just something for the lower-end line I do not know.
The original Spring DAC used an XMOS-based USB input, as did the CYAN, and as far as I can tell as does the current Spring 3 and May, with various USB PHY implementations - though the Spring 3 and May use a different (newer, more powerful) XMOS chipset and overall USB implementation to the original Spring DAC.
Of note … ALL of them seem to have USB isolators. The earlier boards have electrical moats; newer ones (including Spring 3, May, CYAN 2) have electrical and physical moats (board cut-outs) as part of the isolation chip implementation.
… but … again … I’ve not head the CYAN 2 so can’t comment on how it sounds at all, much less on the “quality” of its various input options.
1 Like
All non-true 1-bit DACs have resistor non-linearities. True, parallel, multi-bit designs (inc. R2R) exhibit them to the highest degree, but they are always present at some level. Anything beyond 20-bit isn’t even theoretically audible (the “best” research today suggests 116 dB is the limit, which is between 19 and 20 bits, and properly dithered 16-bit is perceptually equivalent to about 120 dB anyway).
That said …
The Holo Audio Spring and May DACs have, as part of their unique design and selling points, compensated R2R ladders, which are specifically designed to address this non-linearity - so if there were DACs I was going to worry about this with - they would not be it.
Other PGGB Thoughts
I looked at PGGB in mid-2021 (had a nice chat with the author during that process), following a true-blind approach (easy, since can you ABX software to do it as you’re just switching files). Was interesting, but ultimately I didn’t take it further. There were a few reasons for that:
- I could only tell a difference/had a preference for about 1-track in 20 vs. M-Scaler or HQP.
- It required such concentrated, detailed, audition-style, listening to determine a difference in the first place, that it didn’t really matter; it certainly wasn’t this night/day “can’t live without this” difference I’ve heard others describe.
- Wasn’t usable for streaming at all (the algorithms, at the time, need the ENTIRE file, so there wasn’t a quick fix for that, either).
- I would have made my, at-the-time, 50+TB library 64-70x larger. Which would have cost $10s of $1000s to accommodate properly. And there wasn’t a practical way to “only convert the files that benefitted”, as my library is so large now I would be dead before I could listen to every track to make the determination (I have better things to do).
I’m sure things/options and aspects of the processing have changed since then.
5 Likes
Thanks @Torq for describing the R2R dac properties in a very understandable way. It’s good to know that The Spring 3 has compensated R2R ladders and that it may be a non-issue. Once received, I’ll do lots of test tracks to dial in to find what my brain perceives as ‘best’. But only after the Spring has had a chance to settle in.
I don’t advocate for PGGB beyond encouraging people to give it a try. For my ears (brain) it’s very much worth the effort and storage requirements. The PGGB 256 release is miles beyond the early version you tried (in 2021). Processing speed was vastly improved, noise shaping is better and options to dial in to personal preferences have been expanded.
Was the issue you had with streaming over wifi? I stream from my NAS to a desktop Matrix mini-i Pro 4 on a wired connection, using mconnect on my iPhone to play the minim server library to the mini-i Pro 4.
My source library isn’t as large as yours, and the resultant PGGB library is about 42TB. While I haven’t used the RT version of PGGB, it does eliminate the need to write PGGB processed files to disk.
I was able to simplify my chain by getting rid of my MScaler and extra cabling, which I like.
PGGB isn’t for everyone, but it’s worth the bit of effort to use the trial version and reach a decision if it’s a good fit.
No, it was that PGGB could not (and as far as I know still cannot) be used with TIDAL, Qobuz or Apple Music (etc.) as the source - i.e. it is limited to local files only (the algorithm needs the whole file upfront to do its processing).
It does … but then you incur the processing overhead every time to you play a track.
And then, it is more limited than the “offline” version, requires Foobar 2000 (and therefore Windows … which is just not going to happen), thus cannot work with Roon and still cannot handle streaming sources.
It would complicate mine, by adding an additional, dedicated, high-performance PC in the replay chain - which is much more involved (especially being Windows) on an every-day basis that a no-touch/no-update/no-patch black-box.
But … I simplified (and improved) my chain by replacing the streamer->M-Scaler->DAVE chain with a Linn Klimax DSM (Organik).
1 Like